Social Networks and the Decision to Insure

(Cai et al., 2015) · Micro Networks Reading Group

Max Heinze (mheinze@wu.ac.at)

Department of Economics, Vienna University of Economics and Business (WU Vienna)

May 12, 2025

Short Overview

Cai et al. (2015) study how social networks influence the decision by farmers to take up insurance against extreme weather events.

  • They provide two sets of information sessions to farmers,
    • basic ones to one group,
    • and intensive information sessions to another group.
  • They show that for farmers who do not receive the intensive session, having intensively informed friends increases the take-up rate.
  • This effect is driven by knowledge spillovers rather than simple diffusion of purchase decisions.
  • Conducting an additional price randomization experiment on village level enables them to quantify the effect of one additional such friend as being equivalent to a 13 percent reduction in the insurance premium.

Background

  • The study concerns rice farmers in China, i.e., producers of the most important crop in the country.
  • The insurance product is new, meaning that farmers and local officials were unaware of its characteristics.
  • In China, village meetings are a common format of introducing new products or policy and disseminating information about them. The information meetings can be seen in this context.
  • The fair price of the insurance is 12 RMB per mu, but it is heavily subsidized; farmers pay 3.6 RMB, which is deducted from their main subsidy account.
  • Payout is a linear function of the crop loss rate, and maximum payout is 200 RMB. The usual gross income from rice crops is 800 RMB, production cost is 400 RMB.

Experimental Design

In the First Round, households are divided into treated (intensive session) and control (simple session). In the Second Round, farmers are again divided into intensive and simple sessions, and are then shown either no, aggregate, or individualized information on purchasers of the insurance.

What do They Find?

  • Households are asked to list their five closest friends. The share of this group invited to first-round intensive session is then the network measure.
  • The take-up rate in first-round intensive sessions is 14 percentage points higher than in simple sessions (49 vs. 35 percent).
  • Having one additional friend who was treated in the first round increases take-up by 5.8 percentage points, 42 percent of the impact of attending the session oneself.
  • People are not influenced by first-round simple session attendees; they are also less influenced if they are educated about insurance.
  • The price randomization experiment shows that increasing the price by 1 RMB decreases takeup by 12.3 percentage points.
  • Having one additional treated friend is equivalent in take-up effect to a 13-percent decrease in the average insurance premium.

What Is the Effect Driven by?

  • There are three potential channels:
    • People gain knowledge about the value of the product from their informed friends,
    • People learn how to use the product from their friends, and
    • People learn that their friends purchased the product and copy their decisions.
  • Using the accompanying survey, Cai et al. (2015) estimate the knowledge spillover effect and find that people who were informed by friends after the first round have a comparable level of knowledge to their informed friends.
  • As for whether purchase decisions diffuse by themselves, they found that only if information on overall take-up rate or on which friends purchased the product is disseminated to farmers are they influenced by their friends’ decisions — if no information is provided, take-up rate among friends had no effect.

References


Cai, J., Janvry, A. D., & Sadoulet, E. (2015). Social networks and the decision to insure. American Economic Journal: Applied Economics, 7(2), 81–108. https://doi.org/10.1257/app.20130442